# >>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>><<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<
-require 'biopieces'
-require 'fasta'
-require 'pp'
+require 'maasha/biopieces'
+require 'maasha/fasta'
class Velvet
def initialize(directory, sequence_file, verbose)
@verbose = verbose
end
- def run_velveth(kmer_min, kmer_max, type)
- @kmer_min = kmer_min
- @kmer_max = kmer_max
-
- kmer = @kmer_min
-
- while kmer <= @kmer_max
- dir_velveth = [@directory, "Kmer_#{kmer}"].join(File::SEPARATOR)
+ def run_velveth(kmer_min, kmer_max, kmer_step, type)
+ commands = []
+ commands << "velveth"
+ commands << @directory + File::SEPARATOR
+ commands << "#{kmer_min},#{kmer_max + kmer_step},#{kmer_step}" # FIXME - ugly fix for velvet bug
+ commands << "-create_binary"
+ commands << "-#{type}"
+ commands << "-fasta"
+ commands << @sequence_file
+
+ execute(commands)
+ end
- Dir.mkdir(dir_velveth)
+ def run_velvetg(ins_length, paired)
+ dirs = Dir.entries(@directory).select { |d| d.match(/^_/) }
+ dirs.each do |dir|
commands = []
- commands << "velveth"
- commands << dir_velveth
- commands << kmer
- commands << "-#{type}"
- commands << "-fasta"
- commands << @sequence_file
+ commands << "velvetg"
+ commands << File.join(@directory, dir)
+ commands << "-cov_cutoff auto"
+ commands << "-exp_cov auto"
+ commands << "-ins_length #{ins_length}" if paired
+ commands << "-clean yes"
+ commands << "-min_contig_lgth 200"
execute(commands)
-
- kmer += 2
- end
- end
-
- def run_velvetg(cov_cutoffs, exp_cov)
- Dir.glob("#{@directory}/Kmer_*").each do |dir_velveth|
- files_velveth = Dir.glob("#{dir_velveth}/*")
-
- cov_cutoffs.each do |cov_cutoff|
- dir_velvetg = [dir_velveth, "Cov_cutoff_#{cov_cutoff}"].join(File::SEPARATOR)
-
- Dir.mkdir(dir_velvetg)
- FileUtils.cp_r files_velveth, dir_velvetg
-
- commands = []
- commands << "velvetg"
- commands << dir_velvetg
- commands << "-cov_cutoff #{cov_cutoff}"
- commands << "-exp_cov #{exp_cov}"
- commands << "-clean yes"
-
- execute(commands)
- end
end
end
def pick_best_assembly
list = []
- Dir.glob("#{@directory}/Kmer_*/Cov_cutoff_*/contigs.fa").each do |file|
+ Dir.glob("#{@directory}/_*/contigs.fa").each do |file|
n50 = fasta_n50(file)
list << [file, n50]
end
private
def execute(commands)
- commands.unshift "nice -n 19"
commands.push "> /dev/null 2>&1" unless @verbose
command = commands.join(" ")
begin
+ $stderr.puts command if @verbose
system(command)
raise "Command failed: #{command}" unless $?.success?
rescue
def fasta_n50(file)
total = 0
lengths = []
+ count = 0
+ n50 = 0
- Fasta.open(file, mode="r") do |fasta_io|
+ Fasta.open(file, "r") do |fasta_io|
fasta_io.each do |entry|
total += entry.length
lengths << entry.length
end
end
- n50(total, lengths)
- end
-
- def n50(total, lengths)
- count = 0
- n50 = 0
-
lengths.sort.reverse.each do |length|
count += length
end
end
-types = 'short,shortPaired,long,longPaired'
-cov_cutoffs = "2,4,8,16"
+types = 'short,shortPaired,long,longPaired'
casts = []
-casts << {:long=>'directory', :short=>'d', :type=>'dir', :mandatory=>true, :default=>nil, :allowed=>nil, :disallowed=>nil}
-casts << {:long=>'type', :short=>'t', :type=>'string', :mandatory=>true, :default=>'short', :allowed=>types, :disallowed=>nil}
-casts << {:long=>'kmer_min', :short=>'k', :type=>'uint', :mandatory=>true, :default=>19, :allowed=>nil, :disallowed=>nil}
-casts << {:long=>'kmer_max', :short=>'K', :type=>'uint', :mandatory=>true, :default=>31, :allowed=>nil, :disallowed=>nil}
-casts << {:long=>'cov_cutoff', :short=>'c', :type=>'list', :mandatory=>true, :default=>cov_cutoffs, :allowed=>nil, :disallowed=>nil}
-casts << {:long=>'exp_cov', :short=>'e', :type=>'float', :mandatory=>true, :default=>'auto', :allowed=>nil, :disallowed=>nil}
-casts << {:long=>'clean', :short=>'x', :type=>'flag', :mandatory=>false, :default=>nil, :allowed=>nil, :disallowed=>nil}
+casts << {:long=>'directory', :short=>'d', :type=>'dir', :mandatory=>true, :default=>nil, :allowed=>nil, :disallowed=>nil}
+casts << {:long=>'type', :short=>'t', :type=>'string', :mandatory=>true, :default=>'short', :allowed=>types, :disallowed=>nil}
+casts << {:long=>'kmer_min', :short=>'k', :type=>'uint', :mandatory=>true, :default=>19, :allowed=>nil, :disallowed=>nil}
+casts << {:long=>'kmer_max', :short=>'K', :type=>'uint', :mandatory=>true, :default=>31, :allowed=>nil, :disallowed=>nil}
+casts << {:long=>'kmer_step', :short=>'s', :type=>'uint', :mandatory=>true, :default=>2, :allowed=>nil, :disallowed=>nil}
+casts << {:long=>'ins_length', :short=>'i', :type=>'uint', :mandatory=>false, :default=>nil, :allowed=>nil, :disallowed=>nil}
+casts << {:long=>'clean', :short=>'X', :type=>'flag', :mandatory=>false, :default=>nil, :allowed=>nil, :disallowed=>nil}
-bp = Biopieces.new
-
-options = bp.parse(ARGV, casts)
+options = Biopieces.options_parse(ARGV, casts)
raise ArgumentError, "kmer_min #{options[:kmer_min]} must be uneven." if options[:kmer_min].even?
raise ArgumentError, "kmer_max #{options[:kmer_max]} must be uneven." if options[:kmer_max].even?
+raise ArgumentError, "kmer_step #{options[:kmer_step]} must be even." unless options[:kmer_step].even?
raise ArgumentError, "kmer_min >= kmer_max: #{options[:kmer_min]} >= #{options[:kmer_max]}" unless options[:kmer_max] > options[:kmer_min]
Dir.mkdir(options[:directory]) unless Dir.exists?(options[:directory])
-file_fasta = [options[:directory], "sequence_in.fna"].join(File::SEPARATOR)
+file_fasta = File.join(options[:directory], "sequence_in.fna")
-Fasta.open(file_fasta, mode="w") do |fasta_io|
- bp.each_record do |record|
- fasta_io.puts record
- end
-end
-
-unless File.size(file_fasta) == 0
- velvet = Velvet.new(options[:directory], file_fasta, options[:verbose])
- velvet.run_velveth(options[:kmer_min], options[:kmer_max], options[:type])
- velvet.run_velvetg(options[:cov_cutoff], options[:exp_cov])
- file_contigs = velvet.pick_best_assembly
-
- Fasta.open(file_contigs, mode="r") do |fasta_io|
- fasta_io.each do |entry|
- bp.puts entry.to_bp
+Biopieces.open(options[:stream_in], options[:stream_out]) do |input, output|
+ Fasta.open(file_fasta, "w") do |fasta_io|
+ input.each_record do |record|
+ if record[:SEQ_NAME] and record[:SEQ]
+ seq = Seq.new_bp(record)
+ fasta_io.puts seq.to_fasta
+ end
end
- end
+ end
+
+ unless File.size(file_fasta) == 0
+ velvet = Velvet.new(options[:directory], file_fasta, options[:verbose])
+ velvet.run_velveth(options[:kmer_min], options[:kmer_max], options[:kmer_step], options[:type])
+ velvet.run_velvetg(options[:ins_length], options[:type].match("Paired"))
+ file_contigs = velvet.pick_best_assembly
+
+ kmer = file_contigs.match(/_\d+/)
+
+ Fasta.open(file_contigs, "r") do |fasta_io|
+ fasta_io.each do |entry|
+ entry.seq_name << "_kmer#{kmer}"
+ output.puts entry.to_bp
+ end
+ end
+ end
end
+FileUtils.remove_entry_secure file_fasta
FileUtils.remove_entry_secure options[:directory] if options[:clean]
-
# >>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>><<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<
projects / biopieces.git / blobdiff